Arash Hashemi is a research engineer in the motion research group (MoRG), University of Waterloo. Arash specializes in data-driven and intelligent control and their application in robotics and autonomous vehicles. His focus is on integrating reinforcement learning and model-based controllers. He has extensively worked on the real-time implementation of nonlinear model predictive controllers (NMPC) and sliding mode controllers (SMC). In addition, Arash is interested in using various methods of deep learning (vision networks, time series, transformers) for developing accurate and fast high-fidelity/control-oriented models for real-time control. He is currently working on adaptive admittance controllers for rehabilitation robotics, deep-learning-based human-robot interaction models, and mobile deployment of advanced pose-detection networks.
Prior to his current position, Arash completed his master’s degree under the supervision of Prof. John McPhee, Canada research chair in biomechatronics systems dynamics. His research was on automatic weight adjustment of model-based controllers using deep reinforcement learning. He applied multiple real-time controllers to a rehabilitation robot. He also used direct-collocation for rehabilitation robot trajectory planning.
Arash used Lyapunov-based nonlinear controllers in his bachelor's education. He applied sliding mode controller (SMC) on an AFM micro-robot for vibration control.
His research interests are: deep reinforcement learning, deep learning, model-based control theory, robotics, autonomous vehicles
You can find more information about Arash's research on his Linkedin: www.linkedin.com/in/arashhashemi-masc